Establishing AI competency
Snaring the promise of generative AI requires more than executive buy-in and tech investment. The harder task is to build an organization that is skilled enough to experiment, empowered enough to act and guided by boundaries clear enough to trust.
Sometimes, confusion breeds an interest in innovation. Julian Nolan, founder and CEO of Iprova, a Swiss company that uses data to support R&D and intellectual property teams, recalls how recently, “one customer explained that they were talking to us, at least partly, because their board had told them to use more AI in research. The board might not know what AI is, but they do know if they use it, it could be good.” This attitude, he adds, “sums up nicely” where a lot of companies are.
How can I help you today?
- Which of your decisions are just expensive pattern-matching?
- Where does your edge depend on knowing something others don't?
- What breaks when prediction becomes free?
Indeed, using AI effectively is widely seen by executives as today’s bright shiny object. With good reason: Researchers from the Wharton School estimated in a fall 2025 paper that generative AI’s use alone will grow the economy by 1.5% over the next decade. But it won’t make a difference at companies where befuddled corporate leaders simply tell management and employees to “make it so” and hope for the best. To share in the imminent prize, leaders have to understand the technology’s possibilities and risks, assess what specific changes they may want to focus on when using it – and then determine which types of skills and tools need to be in place for widespread adoption. Developing the competency to deploy AI in every part of the business requires deliberate action combined with openness to organic change. To come out on the winning side of this upheaval, the C-Suite must promote and create a corporate culture of experimentation that encourages, within clear guardrails, everyone across the organization to use AI.
Yet this future is struggling to appear at many firms. Just look at hotly contested research from the Massachusetts Institute of Technology’s (MIT) Networked AI Agents in Decentralized Architecture project that assessed the success rate of GenAI experiments. It raised heckles and rebuttals when it reported in July 2025 that 95% of generative AI pilot projects fail – only the technology and media sectors seeing any signs of competitive disruption. Slow progress to date does not surprise Nolan. “New technologies,” he says, “always have much longer gestation cycles than people expect.” Yet this shouldn’t reassure executives as delays in technology adoption are usually a collective calm before a huge storm.
Already visible changes point to a universal challenge. Bob Goodson – the founder and president of Quid, which uses AI to deliver customer context to major companies sees “some roles changing pretty massively.” Those changes will only multiply as investment in pilot projects is widespread, according to the MIT study. And, as Goodson points out, “The company that figures out how to use AI to reduce costs or drive revenue is going to have a meaningful advantage in many fields.” Even if only 5% of experiments currently yield results, he adds, the businesses involved are set to make big gains. The key to joining that select group is to adopt a wider lens that looks beyond AI. “This is not really a technology question,” says Nolan. “It is more about the ability of humans to adapt.”
Simple change management strategies will be of limited use. Unlike for every new kind of software going back decades, says Goodson, “AI has no instruction manual.” Beyond the mushrooming number of widely available business tools, “we still don’t know what large language models (LLMs) can do. We’re in the Wild West of figuring that out.” As a result, targeted deployment strategies which have worked before, even amid previous technological disruptions, won’t work.
In this environment, says Melissa Cheals, CEO of Smartly, a New Zealand payroll and HR services company, “you need really curious people open to putting experimentation right at the heart of your business and a culture that encourages and supports its use. However, you also need to have effective guardrails in place to ensure safe use.” Everything from large pilot studies to individuals just trying something different becomes basic to progress. Instilling across the company the necessary human competency to engage in such activity and take full advantage of AI will demand executive engagement on what to do and where to focus, wider cultural changes and the spread of experience-driven learning across the company.
In this environment, says Melissa Cheals, CEO of Smartly, a New Zealand payroll and HR services company, “you need really curious people open to putting experimentation right at the heart of your business and a culture that encourages and supports its use. However, you also need to have effective guardrails in place to ensure safe use.” Everything from large pilot studies to individuals just trying something different becomes basic to progress. Instilling across the company the necessary human competency to engage in such activity and take full advantage of AI will demand executive engagement on what to do and where to focus, wider cultural changes and the spread of experience-driven learning across the company.
The to-do list for boards and senior executives is complicated and multifaceted. It begins with creating a company ready to benefit from AI. As Cheals notes, the technology’s introduction can upend operating models: “I’ve heard it described as open-heart surgery.” So it makes sense to create a “GenAI Playbook for Organizations,” as Andy Wu argues. The associate professor of business administration at Harvard Business School co-authored an article by that name in late 2025. In an interview, he lists steps that board members and other senior executives will need to take, independent of AI itself. “The biggest mistake is companies rushing to buy a ChatGPT license when they first need to get their tech debt in order.” He adds, “most boards have a very poor understanding of the state of their IT infrastructure, and most companies and their employees are ill equipped to effectively use any generative AI tools.” Leaders need to understand which company data can yield competitive advantage, where it is, and eliminate storage silos and other barriers to access.
Another foundational competency for the board and corporate leaders is understanding which practices need fundamental rewiring to take full advantage of generative AI. Wu gives two of many possible examples. First, he says, “current incentives and KPIs in organizations are based on a pre-AI world” which will not reward what has become possible. For instance, legal departments currently incentivize a lack of mistakes in written contracts. Because AI greatly increases the speed of contract writing, companies might benefit more “to incentivize throughput with an allowance for error.”
Leadership development will also have to change. Much of the tedious work increasingly given to AI once helped train and season new executive-track hires. “One possible model for the future will be to just throw junior people into high-level tasks that we would previously have thought of as too risky to trust them with,” says Wu.
Looking directly at company-wide AI adoption, as with any large business transformation, success depends on somebody senior taking charge. Various options exist. Goodson calls the appointment of a single individual, answerable directly to the CEO, “one of the most successful things I’ve seen companies do.” He frequently advises company executives about how to handle the human side of AI adoption and always cautions against this role being folded into a single department, given how each function is likely to use AI differently. At Smartly, Cheals says that combining oversight of AI adoption and data has worked because data is so integral to AI. The key is designating a leader with a clear AI focus. Such executives need their own particular competencies in shaping AI policies and promoting its use both for major AI projects as well as allowing all employees to adopt it into their everyday workflows.
The former begins with knowing how to align AI adoption with the company’s best interests. As Goodson explains, the technology affects every part of the business. Because such a rich capacity opens up so many possibilities in a world of limited resources, executives need to focus when considering major programs. Goodson was already working on generative AI projects nearly a decade before ChatGPT first grabbed headlines. During that time, he says, he learned a fundamental need for success: never be distracted by the technology. Instead, in planning projects, “it’s very important to be ruthlessly outcomes-driven. Think business problem first, and then put together the right people, teams and processes to solve it.” Otherwise, companies will spend a huge amount of money on technology, “and it doesn’t end up doing anything for them.”
The former begins with knowing how to align AI adoption with the company’s best interests. As Goodson explains, the technology affects every part of the business. Because such a rich capacity opens up so many possibilities in a world of limited resources, executives need to focus when considering major programs. Goodson was already working on generative AI projects nearly a decade before ChatGPT first grabbed headlines. During that time, he says, he learned a fundamental need for success: never be distracted by the technology. Instead, in planning projects, “it’s very important to be ruthlessly outcomes-driven. Think business problem first, and then put together the right people, teams and processes to solve it.” Otherwise, companies will spend a huge amount of money on technology, “and it doesn’t end up doing anything for them.”
The search for the outcomes of most interest obviously involves understanding where AI can yield the greatest business benefits. But it also means knowing the limits of what not to do. Iprova, for instance, helps companies with invention creation where valuable patents are a common client goal. But a valuable invention requires a significant element of human ingenuity. Therefore, an LLM that creates things from scratch would be much less valuable than one that simply helps. The technology, Nolan explains, has “to give the right level of assistance – the right information at the right time – in accordance with patent office guidelines.”
More generally, Nolan adds “your AI invention strategy has to be consistent with your ability to act on the outcomes that it is likely to generate.” A recent Harvard Business Review article, which he co-wrote, explains that the sweet spots are those areas where companies have some control over a product’s wider value chain and over the technology used to produce and deliver it. There, AI-driven changes and insights can rapidly be scaled up. In other areas, even good ideas might fall flat.
Setting up rules around AI use is another crucial area where leadership must develop competency fast. Ultimately, the inherent dangers of misusing the technology make this a board responsibility. They absolutely must get this right, Cheals explains. “Without proper governance, AI does pose a risk – particularly when it comes to security and data privacy. This is why it is crucial to ensure the right policies and frameworks are in place.” To do this well means wrestling with some important questions.
“It’s very important to be ruthlessly outcomes-driven. Think business problem first, and then put together the right people, teams and processes to solve it.”
“You need people open to putting experimentation right at the heart of your business and a culture that encourages and supports its use.”
No shortcuts exist to build ai competency across an entire workforce. Simply using the technology to replace humans doesn’t work, as a recent survey by organizational planning platform Orgvue highlights. While 39% of respondents reported cutting jobs in light of AI adoption, 55% of this group now regret the step. Cheals explains why such remorse is no surprise: “AI is not about replacing people, it is about augmenting and enhancing capability. It’s not a machine doing something on its own, it is a tool that still needs to be operated, trained and managed by a person.” Organizations therefore need a focused effort that ensures their people are excited about what technology can do. Once on board, employees will need general training such as learning prompting skills and how to integrate company data into answers from LLMs. Beyond such basics, competency will ultimately come from the experience of playing with and using new tools in their own jobs.
Inculcating AI skills across the whole workforce requires significant shifts in corporate culture. To begin with, encouraging messages from the top can’t be empty words. As Goodson puts it, CEOs “need to be exemplary in their own use.” If leaders play the role model in chief by using the technology and show their teams what they’ve accomplished, it sends a powerful message. The hardest part here, Wu warns, will not be for board members and CEOs to learn how to use generative AI, which he calls “perhaps the easiest technology to get started using of all time.” Instead, they need to unlearn old habits of how to interact with technology.
Next, Nolan stresses, people need to want to adopt AI because it makes them feel empowered in their own roles. “They mustn’t feel threatened,” he explains. “It has to create value for them personally, ideally quite quickly.” Emotional benefits are integral to successful adoption, he adds. If implemented well, AI’s ability to make employees “feel like they’ve got superpowers, and to make their lives more fun and enjoyable, is an incredibly important aspect of getting the human element right.” Before they get too heroic, though, employees also need to gain an understanding of when such experimentation is safe and appropriate. While guardrails are needed, employees need freedom to experiment and learn, which requires managers to be understanding when some efforts inevitably fail, says Wu.
Next, Nolan stresses, people need to want to adopt AI because it makes them feel empowered in their own roles. “They mustn’t feel threatened,” he explains. “It has to create value for them personally, ideally quite quickly.” Emotional benefits are integral to successful adoption, he adds. If implemented well, AI’s ability to make employees “feel like they’ve got superpowers, and to make their lives more fun and enjoyable, is an incredibly important aspect of getting the human element right.” Before they get too heroic, though, employees also need to gain an understanding of when such experimentation is safe and appropriate. While guardrails are needed, employees need freedom to experiment and learn, which requires managers to be understanding when some efforts inevitably fail, says Wu.
Next, Nolan stresses, people need to want to adopt AI because it makes them feel empowered in their own roles. “They mustn’t feel threatened,” he explains. “It has to create value for them personally, ideally quite quickly.” Emotional benefits are integral to successful adoption, he adds. If implemented well, AI’s ability to make employees “feel like they’ve got superpowers, and to make their lives more fun and enjoyable, is an incredibly important aspect of getting the human element right.” Before they get too heroic, though, employees also need to gain an understanding of when such experimentation is safe and appropriate. While guardrails are needed, employees need freedom to experiment and learn, which requires managers to be understanding when some efforts inevitably fail, says Wu.
Experimental learning shouldn’t be a solo pursuit either. At Smartly, generative AI training is mostly an interactive group activity by design. “If you’re all together – and this can be anybody and everybody in the company – then you’re able to share and see what other people have done,” explains Cheals. In such an environment, she says, the possibilities become real for every employee in their daily jobs. Wu adds that this kind of interaction lets companies tap into the knowledge of those employees who are already fully competent in GenAI use and spread it more broadly. Often, he says, these are the youngest employees, especially those who were still in college or university when ChatGPT was released. They tend to be “absolute ninjas at using ChatGPT and generative AI tools,” says Wu. “A lot of this knowledge is trapped at the bottom. It needs to be shared upward through the organization.”
When done correctly, collective experimentation can even blur the line between training and helping the company. Cheals recalls ”a couple of fantastic online brainstorming sessions” which led to a list of 40 potentially transformational product developments. She describes the next stage as “crazy sessions” because they let participants work with generative AI to gain experience in unfamiliar territory. Participants used AI to prototype new tools, which allowed them to rapidly turn some of the top identified ideas into functional products. One such tool, now live, allows Smartly clients to automatically generate organizational HR charts, which are based on how relationships currently work within their company. Another upcoming tool will allow customers to provide new hires with automated, self-serve onboarding.
Organizations embarking on this journey will be walking a fine line. Encouraging everyone to put GenAI to use while the understanding of what the technology can do is still evolving may make it hard to look at KPIs and other metrics to determine if the organization is on the right track. But if executives and employees are clearly excited about venturing outside their comfort zone with AI by their side, then the metrics will follow the momentum.
-
Use AI with purpose, not pressure:
Organizations see AI as strategic, but real value comes from solving business problems — not adopting it because leadership expects it.
-
Build culture before scale:
Most AI initiatives struggle because organizations lack the infrastructure, governance and culture needed for effective adoption.
-
Create space for experimentation:
Companies benefit most when employees can test and learn from AI within clear frameworks balancing innovation, security and privacy.
Stay ahead of the curve!
Sign up for our newsletter and let Think:Act bring you up to speed with what is happening today and guide you on what’s happening next.
.svg)
